Application of the MODIS global supervised classification model to vegetation and land cover mapping of Central America

While mapping vegetation and land cover using remotely sensed data has a ri ch history of application at local scales, it is only recently that the cap ability has evolved to allow the application of classification models at re gional, continental and global scales. The development of a comprehensive t raining, testing and validation site network for the globe to support super vised and unsupervised classification models is fraught with problems impos ed by scale, bioclimatic representativeness of the sites, availability of a ncillary map and high spatial resolution remote sensing data, landscape het erogeneity, and vegetation variability. The System for Terrestrial Ecosyste m Parameterization (STEP)-a model for characterizing site biophysical, vege tation and landscape parameters to be used for algorithm training and testi ng and validation-has been developed to support supervised land cover mappi ng. This system was applied in Central America using two classification sys tems based on 428 sites. The results indicate that: (1) it is possible to g enerate site data efficiently at the regional scale; (2) implementation of a supervised model using artificial neural network and decision tree classi fication algorithms is feasible at the regional level with classification a ccuracies of 75-88%; and (3) the STEP site parameter model is effective for generating multiple classification systems and thus supporting the develop ment of global surface biophysical parameters.